The SageMaker Python SDK permits an authenticated remote user to read the HMAC signing key from responses of the DescribePipeline API. The key is stored in cleartext within the @step and @remote decorator’s pipeline components prior to SDK versions 3.11.0 and 2.256.0. An attacker can then forge valid integrity signatures for crafted function payloads, enabling execution of arbitrary code within another user's pipeline execution context inside the same AWS account.

The vulnerability affects AWS SageMaker Python SDK versions older than 3.11.0 and 2.256.0. Users who rely on the @step or @remote pipeline decorators in these releases are at risk.

The CVSS score is 8.5, indicating high severity. EPSS information is not available, and the vulnerability is not listed in the CISA KEV catalog. Attack requires an authenticated user with access to the same AWS account; it can be exploited by impersonating the target during pipeline execution.